Skip to content

Issues: NVIDIA/nccl

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

686 Open 707 Closed
686 Open 707 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Potential bug in group.cc
#1599 opened Feb 8, 2025 by songhexiang
1
why NCCL_MAX_NCHANNELS cannot limit ncclDevKernel_SendRecv grid size
#1598 opened Feb 8, 2025 by Graham1025
Is there a relationship between ncclTopoSearchNextGpuSort and followPath
#1596 opened Feb 7, 2025 by zhangdexin
Question about algorithms used for all-gathers and reduce-scatters
#1594 opened Feb 5, 2025 by siddharth9820
4
Design question: what is comm->planner.tmpCollWorkQueue used for?
#1592 opened Jan 30, 2025 by YconquestY
NCCL internal error for ncclCommInitRank when using infiniband
#1591 opened Jan 29, 2025 by SzymonOzog
5
Error while attaching to shared memory segment
#1589 opened Jan 25, 2025 by cassanof
1
In the scenario with two H20 nodes, each equipped with 16 GPU cards and 4 InfiniBand (IB) network cards per node, the bus bandwidth is only reaching 65Gbps.
#1588 opened Jan 24, 2025 by PaggyZhang
2
Potential buffer overflow issue in xml reader
#1585 opened Jan 21, 2025 by fishautumn
Getting NVLS user buffer registration to work under PyTorch
#1584 opened Jan 20, 2025 by lw
1
what does device network offload mean?
#1582 opened Jan 20, 2025 by kkndyu
NCCL Ignores Specified SOCKET_IFNAME Configuration on Worker Nodes in Multi-Node Setup
#1581 opened Jan 18, 2025 by rachid2198
5
NCCL_SOCKET_IFNAME has no effect during pytorch distributed training with multiple NICs
#1580 opened Jan 18, 2025 by hanruijiang
10
Question about post recv
#1577 opened Jan 17, 2025 by lpoyster
BusBW of 2-node tree-based Allreduce exceeds the theoretical limit
#1576 opened Jan 16, 2025 by JK-Jiagn
[Q] How to build nvidia-nccl whl package?
#1575 opened Jan 15, 2025 by kehuanfeng
nccl connection abort(between kubernetes pods): WARN NET/IB: read failed in ncclIbRoceGetVersionNum: Invalid argument
#1573 opened Jan 10, 2025 by limu713
8
nccl_max_nchannels dynamically set in the program
#1572 opened Jan 9, 2025 by Rainlin007
2
net.cc recvProxyConnect p2p buff set
#1570 opened Jan 9, 2025 by GeofferyGeng
Potential group\collective life-time management issue in profiler plugin.
#1569 opened Jan 9, 2025 by wiryls
2
How does the GPU that initiates the nvswitch broadcast know that the nvswitch has completed the broadcast operation?
#1568 opened Jan 8, 2025 by pengqianqian
Running issue with Nvidia NIM for NCCL.
#1567 opened Jan 7, 2025 by innat-asj
4
group launch allreduce + allgather
#1566 opened Jan 7, 2025 by alpha-baby
1
Comment of function
#1565 opened Jan 6, 2025 by WoosungMyung
1
[Hopper/NVLINK4] Origin of failure of fabric manager manifested through NCCL-based codes
#1562 opened Jan 3, 2025 by vitduck
3
ProTip! Type g i on any issue or pull request to go back to the issue listing page.